vault backup: 2023-11-09 14:26:01

.obsidian/workspace.json

@@ -41,7 +41,7 @@
             "state": {
               "type": "markdown",
               "state": {
-                "file": "05. 資料收集/Keras.tensorflow - Dataset.md",
+                "file": "05. 資料收集/Keras.tensorflow - shuffle.md",
                 "mode": "source",
                 "source": true
               }
@@ -114,7 +114,7 @@
             "state": {
               "type": "backlink",
               "state": {
-                "file": "05. 資料收集/Keras.tensorflow - Dataset.md",
+                "file": "05. 資料收集/Keras.tensorflow - shuffle.md",
                 "collapseAll": false,
                 "extraContext": false,
                 "sortOrder": "alphabetical",
@@ -139,7 +139,7 @@
             "state": {
               "type": "outline",
               "state": {
-                "file": "05. 資料收集/Keras.tensorflow - Dataset.md"
+                "file": "05. 資料收集/Keras.tensorflow - shuffle.md"
               }
             }
           },
@@ -167,7 +167,7 @@
             "state": {
               "type": "file-properties",
               "state": {
-                "file": "05. 資料收集/Keras.tensorflow - Dataset.md"
+                "file": "05. 資料收集/Keras.tensorflow - shuffle.md"
               }
             }
           }
@@ -212,6 +212,8 @@
   },
   "active": "9ef635642faa0c82",
   "lastOpenFiles": [
+    "05. 資料收集/Keras.tensorflow - Dataset.md",
+    "05. 資料收集/Keras.tensorflow - shuffle.md",
     "01. 個人/01. Daily/2023-11-08(週三).md",
     "01. 個人/01. Daily/2023-11-07(週二).md",
     "01. 個人/01. Daily/2023-11-06(週一).md",
@@ -238,11 +240,9 @@
     "01. 個人/01. Daily/2023/05",
     "01. 個人/01. Daily/2023-10-25(週三).md",
     "00. Inbox/My Mindmap.canvas",
-    "05. 資料收集/Keras.tensorflow - Dataset.md",
     "05. 資料收集/AI.md",
     "00. Inbox/Habit Tracker.md",
     "00. Inbox/台語諺語.md",
-    "05. 資料收集/99. templates/日記.md",
     "00. Inbox/vc-fwUpdate code trace(Meetup).canvas",
     "05. 資料收集/Tool Setup/Software/diskstation/share/Tools/字型",
     "05. 資料收集/Tool Setup/Software/diskstation/share/Tools",

05. 資料收集/Keras.tensorflow - shuffle.md

@@ -0,0 +1,34 @@
+如果想用同時打亂x_train與y_train，可以參考這2個方法。
+
+## 1. 用 `tf.random.shuffle`
+```python
+indices = tf.range(start=0, limit=tf.shape(x_data)[0], dtype=tf.int32)
+idx = tf.random.shuffle(indices)
+x_data = tf.gather(x_data, idx)
+y_data = tf.gather(y_data, idx)
+```
+
+先建立一個跟array一樣大的list，然後打亂它，再用這個已打亂的list當作索引來建立一個新的data list。
+
+## 2. 
+```python
+x_train = tf.data.Dataset.from_tensor_slices(x)
+y_train = tf.data.Dataset.from_tensor_slices(y)
+
+x_train, y_train = x_train.shuffle(buffer_size=2, seed=2), y_train.shuffle(buffer_size=2, seed=2)
+dataset = tf.data.Dataset.from_tensor_slices((x_train, y_train))
+```
+
+或者
+
+```python
+BF = 2
+SEED = 2
+def shuffling(dataset, bf, seed_number):
+   return dataset.shuffle(buffer_size=bf, seed=seed_number)
+
+x_train, y_train = shuffling(x_train, BF, SEED), shuffling(y_train, BF, SEED)
+dataset = tf.data.Dataset.from_tensor_slices((x_train, y_train))
+```
+
+概念跟第一點是一樣的，但是這是先轉成 `tf.data.Dataset`，然後把x_train跟y_train都用同樣的seed打亂。